Fractional CTO for startups

I help startups that need senior technical leadership but not a full-time CTO yet. I set the technology strategy, design the architecture, hire and lead the team, and I still write the code myself.

You need this if

• You are a non-technical founder and need someone to own the technology.
• You have a few developers but nobody setting direction, standards and priorities.
• You are about to raise money or hire, and the technical side has to hold up.

What I do

• Strategy: technology choices, roadmap, build or buy, tied to your business goals.
• Architecture: systems designed for the problem you have: serverless, microservices or a well-structured monolith.
• Team: hiring, onboarding, code review and engineering practices.
• Hands-on code: Go and Elixir on AWS, with infrastructure as code.

Architecture review and team training

When the team is there but the architecture is getting complex, I review it and train the developers. In April 2026, at Levels, the Milan startup behind Operatore Sicurezza, I reviewed the technical architecture, trained the team on best practices for complex architectures, and helped them choose the right technologies and fix the weak spots.

MVP from zero to production

I take a product from idea to production: scope, architecture, code, infrastructure and deployment. I have shipped 6 products from zero to production, the latest travelogue.social at Travelware. At brumbrum the first production system went live in 9 months, and file.dentist I designed and built alone, from product design to marketing.

AWS scaling and rescue

When the architecture does not hold anymore, or the AWS bill grows faster than the business, I redesign it: event-driven systems, serverless where it fits, CQRS and event sourcing where they pay off, infrastructure as code with CDK. At brumbrum I ran 120 microservices serving 10M+ requests/day on less than €100k/year. At Macai, order processing ran with sub-200ms p99 latency.

AI for companies

I build AI into products and into the way companies work: MCP servers that connect AI assistants to your systems, Claude Code with skills, agents and automations, and RAG over company data. I have built a semantic search engine from scratch with vector embeddings (AWS Bedrock, Qdrant) and real-time NLP pipelines for sentiment and impact analysis.

How we work together

Part-time, with the time commitment that fits your stage. Remote across EMEA or on site in Milan, in Italian or English.